INDEX

Explanations

The neuron fires strongly on interface or document‐structure labels and calls to action for downloadable or viewable resources—e.g. “Download,” “Full screen,” “Information,” “Description,” “Download mp3,” “Rubric,” “Image gallery,” and so on.

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

Fragmentation

0.37

precio

0.34

correo

0.34

ጣል

0.34

ፈላ

0.32

Pourquoi

0.32

PROBLE

0.32

honneur

0.32

bestroute

0.31

Naissance

0.31

POSITIVE LOGITS

and

0.51

for

0.51

 with

0.48

 from

0.43

0.41

＆

0.39

0.37

 для

0.37

by

0.36

 worksheet

0.35

Activations Density 3.236%